Hai Family

The problem Families run on shared plans and personal secrets at the same time. Someone’s birthday is coming up, one sibling is planning a surprise party, a parent has a private reminder, and grandparents need to know about Thanksgiving. Today that coordination is spread across group chats, calendars, and people’s heads. Things get missed, or surprises get spoiled. AI assistants could help, but they’re built for one person. If a whole family shares one assistant, it learns everyone’s information, and there is nothing to stop it repeating one person’s secret to another. Telling the model “don’t share this” in a prompt isn’t a real safeguard: a cleverly worded question or a model mistake can still leak it. Without real privacy, a family can’t trust a shared AI with anything that matters. The solution Hai Family is one AI companion that every family member chats with privately. It keeps track of birthdays, plans, and family memories, and it always knows who is allowed to see what. Smart reminders with the right audience. Hai brings up what needs attention, like an upcoming birthday, and decides who should hear about it. Private events go only to their owner, secrets go only to the people who are in on them, and birthday reminders go to everyone except the birthday person. Memories with clear visibility. When someone tells Hai something worth remembering, Hai works out who is involved, the date, and whether it’s private, shared with the family, or a secret. Before anything is shared with the whole family, Hai asks the person to confirm the wording and who can see it. Privacy enforced in code, not in the prompt. Every memory passes a permission check before the model sees it. A shared memory is visible to everyone. Anything else is visible only to its owner or to people the owner explicitly shares it with. The model can’t leak what it never received. Saving a shared memory without confirmation is blocked by the program itself, not left to the model’s judgment.